1. Field of the Invention
The present invention relates to a non-coherent detection apparatus and method, and more particularly, to a non-coherent detection apparatus and method for an IEEE 802.15.4 low-rate wireless personal area network (LR-WPAN) binary phase shift keying (BPSK) receiver.
2. Description of the Related Art
A low-rate wireless personal area network (LR-WPAN) defined in an IEEE 802.15.4 standard has been developed for a small amount of processing and low costs, use of small energy, and local area wireless communication in cheap and mobile devices. The IEEE 802.15.4 LR-WPAN standard is widely used in home automation and home appliance markets in which home appliances operate at a maximum bit rate of 10 kbps or less. A 868/915 MHz binary phase shift keying (BPSK) physical layer from among four physical layers of the IEEE 802.15.4 standard supports a maximum bit rate of 40 kbps.
When a receiver having low power consumption and complexity is implemented, a non-coherent detection method by which coherence of a carrier wave required is used. Recently, a demodulation method using non-coherent detection by which a self-correlation method is applied to differential detection of bitstreams by using approximation has been suggested in the IEEE 802.15.4 LR-WPAN BPSK. However, it is assumed that symbol timing that is not easily obtained by the self-correlation method is known in the demodulation method. This is because, in the self-correlation method, a flat timing metric is constituted in a preamble portion of a received signal. Furthermore, the self-correlation method does not distinguish spread spectrum symbols. In other words, the self-correlation method works equally well for similar spread spectrum signals with different spreading codes.
FIG. 1 is a block diagram of a transmission signal model of an IEEE 802.15.4 LR-WPAN BPSK system. Referring to FIG. 1, a PHY protocol data unit (PPDU) bit sequence E[m] is differentially modulated. Each of modulated bits is allocated by a pseudorandom number (PN) C. When each modulated bit is 0, it is allocated by C, and in other cases, it is allocated by 1-C. In this case, the length of C satisfies N=15. PN sequences for the successive bits are concatenated, and the aggregate chip sequence is modulated by BPSK and then is raised-cosine pulse shaped. The raised-cosine pulse shaped chip sequence is up-converted and then is propagated through a radio frequency (RF) transmission module. A chip rate is 300 kChip/s at a band of 868 MHz and 600 kChip/s at a band of 915 MHz. In this case, a signal r[n] of a baseband received at time n is expressed using Equation 1:r[n]=s[n]ej(ω0nTc+θ)+ηo[n]  (1)where s[n] is a BPSK-modulated chip sequence at a transmission terminal, ω0 is a carrier wave frequency offset, θ is a phase offset, Tc is a duration of a chip, and η0[n] is a complex AWGN sequence having a power of 1/signal-to-noise-ratio (SNR).
Meanwhile, there are two non-coherent detection methods. One is chip-level detection, and the other one is bit-level detection. In chip-level detection, a BPSK-modulated chip is searched for by using a phase difference between adjacent chips, as expressed in Equation 2:
                                          c            ⁡                          [              n              ]                                =                                                    r                ⁡                                  [                  n                  ]                                            ⁢                                                r                  *                                ⁡                                  [                                      n                    -                    1                                    ]                                                      =                                                            s                  ⁡                                      [                    n                    ]                                                  ⁢                                  s                  ⁡                                      [                                          n                      -                      1                                        ]                                                  ⁢                                  ⅇ                                                            jω                      0                                        ⁢                                          T                      c                                                                                  +                                                η                  1                                ⁡                                  [                  n                  ]                                                                    ,                                  ⁢                                            s              ^                        ⁡                          [              n              ]                                =                      {                                                                                                                                                        s                          ^                                                ⁡                                                  [                                                      n                            -                            1                                                    ]                                                                    ,                                                                                                                          Re                        ⁢                                                  {                                                      c                            ⁡                                                          [                              n                              ]                                                                                }                                                                    ≥                      0                                                                                                                                                          -                                                                              s                            ^                                                    ⁡                                                      [                                                          n                              -                              1                                                        ]                                                                                              ,                                                                                                                          Re                        ⁢                                                  {                                                      c                            ⁡                                                          [                              n                              ]                                                                                }                                                                    <                      0                                                                                  ,                                                          (        2        )            where (•)* is a complex conjugate operator, η1[n] is a noise term, ŝ[n] is a chip that is detected at time n, and Re{x} is a real-number part of x.
Chip-level detection expressed by Equation 2 may be comparatively easily implemented. However, a detection performance may be lowered, and in particular, when a frequency offset is high, the detection performance may be lowered.
A significantly higher detection performance than in chip-level detection is obtained in bit-level detection, because a soft decision value of a chip is synthesized for bit detection and a frequency offset effect is compensated for. In this case, assuming that an m-th bit of a PPDU starts from time Nm, in bit-level detection, the m-th bit of the PPDU is detected using Equation 3:
                                                                                          A                  0                                ⁡                                  [                  m                  ]                                            =                            ⁢                                                z                  0                                ⁡                                  [                                                                                    (                                                  m                          +                          1                                                )                                            ⁢                      N                                        -                    1                                    ]                                                                                                        =                            ⁢                                                ∑                                      k                    =                    0                                                        N                    -                    1                                                  ⁢                                                      r                    ⁡                                          [                                              Nm                        +                        k                                            ]                                                        ⁢                                                            r                      *                                        ⁡                                          [                                                                        N                          ⁡                                                      (                                                          m                              -                              1                                                        )                                                                          +                        k                                            ]                                                                                                                                              =                            ⁢                              {                                                                                                                                                                                        N                              ⁢                                                                                                                          ⁢                                                              ⅇ                                                                  j                                  ⁢                                                                                                                                          ⁢                                  N                                  ⁢                                                                                                                                          ⁢                                                                      ω                                    0                                                                    ⁢                                                                      T                                    c                                                                                                                                                        +                                                          η                              2                                                                                ,                                                                                                                                                  E                            ⁡                                                          [                              m                              ]                                                                                =                          0                                                                                                                                                                                                                                                        -                                N                                                            ⁢                                                                                                                          ⁢                                                              ⅇ                                                                  j                                  ⁢                                                                                                                                          ⁢                                  N                                  ⁢                                                                                                                                          ⁢                                                                      ω                                    0                                                                    ⁢                                                                      T                                    c                                                                                                                                                        +                                                          η                              2                                                                                ,                                                                                                                                                  E                            ⁡                                                          [                              m                              ]                                                                                ≠                          0                                                                                                      ,                                                                                        (        3        )            where η2 is a noise term.
In Equation 3, z0[n] is obtained by a correlator disposed in a bit-level detection apparatus shown in FIG. 2 by using the self-correlation method.
FIG. 2 is a block diagram of a correlator 200 disposed in a conventional bit-level detection apparatus. Referring to FIG. 2, the correlator 200 of the conventional bit-level detection apparatus includes a delay unit 210, a complex conjugate value calculator 220, a multiplier 230, a buffer 240, and a summing-up unit 250. The delay unit 210 delays a received signal r[n] by an N chip. The complex conjugate value calculator 220 calculates a complex conjugate value of the delayed signal and outputs the result of calculation. The multiplier 230 multiplies the received signal r[n] by an output signal of the complex conjugate value calculator 220, and output values of the multiplier 230 are stored in the buffer 240 in units of chip. The summing-up unit 250 sums up the output values of the multiplier 230 stored in the buffer 240 in units of chip and calculates a self-correlation signal z0[n]. In the conventional bit-level detection apparatus including the correlator 200 shown in FIG. 2, the frequency offset effect is compensated for by estimating a whole preamble of the received signal r[n] by using Equation 4:
                                                                        Y                0                            =                            ⁢                                                1                                                            (                                              J                        -                        1                                            )                                        ⁢                    N                                                  ⁢                                                      ∑                                          m                      =                      1                                                              J                      -                      1                                                        ⁢                                                            ∑                                              k                        =                        0                                                                    N                        -                        1                                                              ⁢                                                                  p                        ⁡                                                  [                                                      Nm                            +                            k                                                    ]                                                                    ⁢                                                                        p                          *                                                ⁡                                                  [                                                                                    N                              ⁡                                                              (                                                                  m                                  -                                  1                                                                )                                                                                      +                            k                                                    ]                                                                                                                                                                                                            =                                ⁢                                                      ⅇ                                          j                      ⁢                                                                                          ⁢                      N                      ⁢                                                                                          ⁢                                              ω                        0                                            ⁢                                              T                        c                                                                              +                                      η                    3                                                              ,                                                          (        4        )            
where p[n] is a preamble sequence received at time 0, J is a bit number of a preamble, and η3 is a noise term.
Meanwhile, in a demodulation method using approximation that has been recently suggested, the m-th bit of the PPDU is detected using Equation 5:
                                          E            ^                    ⁡                      [            m            ]                          =                  {                                                                                          0                    ,                                                                                                              Re                      ⁢                                              {                                                                                                            A                              0                                                        ⁡                                                          [                              m                              ]                                                                                ⁢                          q                          ⁢                                                      {                                                          Y                              0                              *                                                        }                                                                          }                                                              ≥                    0                                                                                                                    1                    ,                                                                                                              Re                      ⁢                                              {                                                                                                            A                              0                                                        ⁡                                                          [                              m                              ]                                                                                ⁢                          q                          ⁢                                                      {                                                          Y                              0                              *                                                        }                                                                          }                                                              <                    0                                                                        ,                                              (        5        )            
where q{•} is a four-level quantization function that represents approximation, and the four levels are 1, −1, j, and −j, respectively.
In the bit-level detection method, an excellent detection performance is obtained. However, in the bit-level detection method, it is assumed that symbol timing that is not easily obtained by the self-correlation method is known. This is because, in the bit-level detection method, a flat timing metric is constituted in a preamble portion of a received signal and it is not easily to accurately determine symbol timing. FIGS. 3 and 4 are graphs showing output patterns that are obtained by self-correlation in a preamble portion of a received signal when signal-to-noise ratios (SNRs) are ∞ dB and 3 dB, respectively. Referring to FIGS. 3 and 4, timing synchronization may not be obtained without further consideration. Furthermore, in the demodulation method using self-correlation, the value of a PN code C is not considered. Thus, a spread spectrum signal using a different PN code from that of an IEEE 802.15.4 LR-WPAN BPSK signal may be regarded as the IEEE 802.15.4 LR-WPAN BPSK signal.